Screened poisson surface reconstruction
By M. Kazhdan, Hugues Hoppe. 2013
2,388 citations recorded by the source corpus.
Poisson surface reconstruction creates watertight surfaces from oriented point sets. In this work we extend the technique to explicitly incorporate the points as interpolation constraints. The extension can be interpreted as a generalization of the underlying mathematical framework to a screened Poisson equation. In contrast to other image and geometry processing techniques, the screening term is defined over a sparse set of points rather than over the full domain. We show that these sparse constraints can nonetheless be integrated efficiently. Because the modified linear system retains the same finite-element discretization, the sparsity structure is unchanged, and the system can still be solved using a multigrid approach. Moreover we present several algorithmic improvements that together reduce the time complexity of the solver to linear in the number of points, thereby enabling faster, higher-quality surface reconstructions.
Subfields: Computational Mechanics, Computer Graphics and Computer-Aided Design, 3D Vision & Pose
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computational Mechanics
- Andreas Geiger
- Matthias Nießner
- Michael Niemeyer
- Gerard Pons-Moll
- Jonathan T. Barron
- Yaron Lipman
- Thomas A. Funkhouser
- Leonidas J. Guibas
In Computer Graphics and Computer-Aided Design
- Andreas Geiger
- Jonathan T. Barron
- Ben Mildenhall
- Michael Niemeyer
- Pratul P. Srinivasan
- Matthias Nießner
- Thomas A. Funkhouser
- Songyou Peng
In 3D Vision & Pose
- Andreas Geiger
- Michael Niemeyer
- Matthias Nießner
- Marc Pollefeys
- Yaron Lipman
- Jonathan T. Barron
- Vladlen Koltun
- Songyou Peng
Citations in the corpus
- Occupancy Networks: Learning 3D Reconstruction in Function Space (2018, background): “However, unlike other representations, this approach requires additional non-trivial post-processing steps [3,6,37,38] to generate the final 3D mesh.”
- DeepSDF: Learning Continuous Signed Distance Functions for Shape Representation (2019, methodology): “Classical surface reconstruction methods complete a point cloud into a dense surface by fitting radial basis function (RBF) [11] to approximate implicit surface functions, or by casting the reconstruction from oriented point clouds as a Poisson problem [32].”
- Learning Shape Priors for Single-View 3D Completion and Reconstruction (2018, methodology): “, Poisson surface reconstruction solves an indicator function on a voxel grid via the Poisson equation [29,28].”
- A Papier-Mache Approach to Learning 3D Surface Generation (2018, positive): “Note that our approach can be directly evaluated by Metro while the baseline requires performing PSR [17].”
- Shape Completion Using 3D-Encoder-Predictor CNNs and Shape Synthesis (2016, methodology): “Poisson surface reconstruction [14, 15] is mostly used to obtain complete surfaces on dense point clouds, but it cannot infer missing structures.”
- Deformable Shape Completion with Graph Convolutional Autoencoders (2017, methodology): “Completion guided by geometric priors has been explored, for example Poisson filling [22] and self-similarity [27, 48, 32].”
- A Survey of Surface Reconstruction from Point Clouds (2017, neutral): “The main difference be-tween the approaches is that [KH13] solves the problem via a finite-element multi-grid formulation, whereas [CT11] use finite-differences, due to the complexity in discretizing the Hessian term; In particular, the screened Poisson formulation [KH13] is up to two orders of…”
- SAL: Sign Agnostic Learning of Shapes From Raw Data (2019, methodology): “To approximate the signed distance function, h̄2, we first tried using a state of the art surface reconstruction algorithm [23] to produce watertight manifold surfaces.”
- Neuralangelo: High-Fidelity Neural Surface Reconstruction (2023, methodology): “Follow-up methods typically start with 3D point clouds from multi-view stereo techniques [6, 7, 28, 34] and then perform dense surface reconstruction [13, 14].”
- PGSR: Planar-Based Gaussian Splatting for Efficient and High-Fidelity Surface Reconstruction (2024, neutral): “…commonly used method separates the overall MVS problem into several parts, by initially extracting dense point clouds from multi-view images through block-based matching [37], followed by the construction of surface structures either through triangulation [38] or implicit surface fitting [39], [40].”
- SceneNN: A Scene Meshes Dataset with aNNotations (2016, methodology): “Input PolyMender [14] Poisson [15] Figure 8.”
- MonoSDF: Exploring Monocular Geometric Cues for Neural Implicit Surface Reconstruction (2022, methodology): “5 Next, we use Meshlab’s screened Poisson reconstruction [32] to reconstruct a mesh from point clouds with default parameters.”
- DeepHuman: 3D Human Reconstruction From a Single Image (2019, methodology): “After data capture, we post-process the raw meshes through hole filling [32], remeshing [27] and isolated artifact removal.”
- Shape As Points: A Differentiable Poisson Solver (2021, methodology): “We can differentiably perform point rasterization via inverse trilinear interpolation, similar to the approach in [30,31].”
- ShapeFormer: Transformer-based Shape Completion via Sparse Representation (2022, background): “However, point clouds need to be non-trivially post-processed using classical methods [6, 36, 39, 40] to recover surfaces due to their sparse nature.”
- Deep Projective 3D Semantic Segmentation (2017, neutral): “While other rendering techniques, such as surface reconstruction as in [12], require demanding preprocessing steps of the point cloud in 3D space, splatting could be completely processed in image space.”
- PointCleanNet: Learning to Denoise and Remove Outliers from Dense Point Clouds (2019, background): “In Figure 14, we show the Poisson reconstructions [KH13] and distances from the ground submittedtoCOMPUTERGRAPHICS Forum (7/2019).”
- Neural RGB-D Surface Reconstruction (2021, methodology): “We obtain a mesh by applying screened Poisson surface reconstruction [35] on the resulting point cloud.”